Characterization of two DNA double-stranded break repair-deficient cell lines that express inactive DNA-dependent protein kinase catalytic subunits.

The DNA-dependent protein kinase (DNA-PK) is a trimeric enzyme consisting of a 460-kDa catalytic subunit (DNA-PKcs) and a heterodimeric regulatory complex called Ku, which is comprised of 70 (Ku70) and 86 (Ku80) kDa subunits. Mutations that affect the expression of the catalytic or Ku80 subunits of DNA-PK disrupt both V(D)J recombination and DNA double-stranded break repair pathways. In this report, we show that two previously uncharacterized rodent cell lines that are defective in DNA double-stranded break repair express catalytically inactive DNA-PK. The DNA-PKcs from the DNA double-stranded break repair mutant cell lines IRS-20 and SX-9 assembles on double-stranded DNA but fails to function as a protein kinase. In addition to the kinase defect, the abundance of the DNA-PKcs from both of these cell lines is reduced relative to wild-type controls. These results suggest that the DNA-PKcs gene from each of these cell lines contains mutations that inactivate the enzymatic activity and the expression or stability of the gene product. These data further strengthen the hypothesis that DNA-PK-mediated protein phosphorylation is a necessary component of the DNA double-stranded break repair pathway.

The rejoining of double-stranded DNA breaks induced by ionizing radiation or occurring as intermediates of V(D)J recombination is performed via a biochemical pathway that includes the DNA-dependent protein kinase holoenzyme. DNA-PK 1 is a trimeric complex consisting of a DNA-binding component made up of the 70 and 86 kd subunits of the Ku autoantigen (1, 2) and a catalytic subunit of approximately 460 kDa (3). Cells from the x-ray-sensitive complementation group (xrs)-7, which includes the severe combined immunodeficiency (scid) mouse and the Chinese hamster ovary (CHO) V3 cell line, exhibit reduced expression of the DNA-PKcs, lack measurable DNA-stimulated kinase activity, and are defective for DNA double-stranded break repair and V(D)J recombination (4 -6). Similarly, cells from the xrs-6 complementation group, which contain mutations that reduce the expression of the Ku80 subunit of DNA-PK, exhibit losses of Ku-specific DNA-ending binding activity (7-9) DNA-PK kinase activity (10) and are also defective for DNA double-stranded break repair and V(D)J recombination (11)(12)(13).
Molecular analysis of these DNA-repair mutant cells indicates that DNA-PK is required for the rejoining of doublestranded DNA breaks, but the mechanism by which DNA-PK functions in this process has yet to be elucidated. DNA-PK is a serine and threonine protein kinase that is activated by doublestranded DNA containing single-stranded to double-stranded transitions, such as DNA-ends, nicks, gaps, and stem-loop structures (14). In vitro, the Ku and catalytic subunits of DNA-PK assemble in a DNA-dependent manner (15), and the DNA-bound holoenzyme preferentially phosphorylates substrates that are bound to the same DNA molecule (1,16). DNA-PK has been shown to phosphorylate a broad range of proteins in vitro, most of which are DNA-binding proteins (17), including the Ku70 and Ku80 subunits of DNA-PK (18,19). Based on these data, it has been proposed that DNA-PK binds to double-stranded DNA breaks produced in the cell by DNAdamaging agents or during V(D)J recombination. The DNAbound holoenzyme could then participate in the DNA-rejoining process by phosphorylating Ku and other protein factors that are colocalized with the kinase at the site of the strand breaks.
In this report, we present data that further demonstrates the importance of DNA-PK in the DNA double-stranded break repair process by showing that two DNA double-stranded break repair-deficient cell lines have defects that disrupt the catalytic activity of DNA-PK. The CHO cell lines IRS-20 (20,21) and SX9 (22) both express DNA-PK catalytic subunits that can assemble on double-stranded DNA but lack detectable protein kinase activity. The expression and activity of the Ku subunits of DNA-PK are normal in each cell line, and DNA-PK activity can be restored to both the IRS-20 and SX-9 cell extracts by addition of purified DNA-PKcs. Transfer of human chromosome 8, which contains the DNA-PKcs gene, can also rescue the kinase defect of the IRS-20 cells. These data are consistent with the IRS-20 and SX-9 cells being in the same complementation group as the mouse scid and CHO V3 cells and support the hypothesis that DNA-PK-mediated protein phosphorylation is an essential component of the DNA doublestranded break repair pathway.

EXPERIMENTAL PROCEDURES
Cell Growth Conditions and Cell Extract Preparation-Cells were grown as monolayer cultures (10B2, IRS-20, IRS-20 (Neo8), M10, LX830, SL3147, and H5) (22)(23)(24) in a humidified, 5% CO 2 atmosphere * This work was supported by a Los Alamos National Laboratory LDRD grant, a grant from the U. S. DOE, and National Institutes of Health Grant CA50519. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Immunoblot Analysis-100 g of protein from whole cell extracts was resolved by SDS-PAGE, transferred to nitrocellulose, and probed with the anti-Ku70 mouse monoclonal antibody N3H10 and the anti-DNA-PKcs monoclonal antibodies 18 -2 and 42-26, as described previously (6).
Preparation of Immobilized DNA and Assembly of DNA-Protein Complexes -Sheared salmon sperm DNA dissolved in 10 mM potassium phosphate, pH 8.0, was incubated for 30 min with cyanogen bromideactivated Sepharose CL6-B (Pharmacia Biotech Inc.) at 25°C. The beads were then washed with 100 ml of 1 M ethanolamine, pH 8.0, and then incubated with an additional 40 ml of 1 M ethanolamine for 30 min at 25°C to block the remaining reactive groups on the beads. The DNA-agarose beads were then washed 5 times with 40 ml of deionized water and stored at 4°C prior to use. The amount of DNA coupled to the beads was estimated to be approximately 0.1 mg of DNA/ml of bead, which was calculated by subtracting the amount of DNA that remained after the coupling reaction from the total used in preparing the beads. DNA-protein complexes were formed on the DNA beads by mixing 50 l (250 g total protein) of cell extract with an equal volume of a 1:1 slurry of the DNA-agarose beads in H 2 0. In some cases, as indicated in the Figs. 1C, 1D, 2A, and 4B legends, purified HeLa cell DNA-PKcs or Ku70/80 was used to supplement the mutant cell extracts. The DNAbead reactions were incubated for 30 min at 25°C and then washed three times with 1 ml of ice-cold TM buffer containing 50 mM KCl to remove weakly associated proteins. The washed DNA beads were then processed for immunoblot analysis by boiling with 30 l of SDS sample buffer or used for protein kinase assays as described below.
Purification of DNA-PK-Ku70/80 was purified from HeLa cell nuclear extract by affinity chromatography using an anti-Ku80 antibody matrix. A typical purification was performed using 50 ml of HeLa cell nuclear extract (10 mg/ml) prepared as described previously (16) and a 2-ml anti-Ku80 affinity matrix containing 1 mg/ml Ku80 antibody (25) covalently coupled to protein G-agarose beads. Ku70/80 complexes were adsorbed to the antibody beads by tumbling the HeLa cell extract overnight at 4°C. The column bound Ku70/80 complexes were subsequently washed with HEPES chromatography buffer (50 mM HEPES, pH 7.5, 2 mM EDTA, 0.01% Nonidet P-40, 1 mM dithiothreitol, 20 g/ml phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, pepstatin A, and leupeptin) containing 0.5 M KCl to remove weakly bound proteins. Ku70/80 was eluted from the antibody column using a buffer containing 50 mM Tris-HCl, pH 7.9, 50% ethylene glycol, and 1.75 M MgCl 2 . The affinity purified Ku70/80 was then extensively dialyzed into TM buffer containing 0.1 M KCl prior to use in kinase assays. DNA-PKcs was purified from HeLa cell nuclear extracts as described previously (16) by phosphocellulose P-11, DEAE-Sepharose, heparin-agarose, and Superdex 200 (Pharmacia) column chromatography. Protein concentrations of the DNA-PKcs and Ku70/80 were obtained by Bradford analysis.
Protein Kinase Assays-500 ng of purified recombinant human replication protein A (26) was added to washed DNA-protein complexes assembled on the DNA beads and incubated for 10 min at 25°C. Protein kinase reactions were initiated by addition of 12.5 M ATP containing 5 Ci of [␥-32 P]ATP and incubated for 60 min at 30°C. Kinase reactions were terminated by boiling for 3 min in SDS-PAGE sample buffer, and the phosphorylated proteins were resolved by 10% SDS-PAGE and visualized by autoradiography of the dried gel.

RESULTS
To screen previously uncharacterized DNA double-stranded break repair-deficient cell lines for mutations that affect DNA-PK, we developed an assay to measure the kinase activity of DNA-PK in cell extracts. In this assay, DNA-PK activity is determined by evaluating the phosphorylation of the 32-kDa subunit of recombinant human replication protein A (RPA) (26), which has previously been shown to be a substrate for DNA-PK in vitro (27,28) and may also be phosphorylated by this enzyme in vivo in response to DNA damage (19). To enrich for DNA-PK from cell extracts, protein kinase reactions are first assembled on a DNA substrate covalently attached to agarose beads. This allows for the removal of weakly bound protein kinases by washing the DNA beads prior to initiating the kinase reactions.
We first tested the phosphorylation of RPA by DNA-PK using these assay conditions with purified HeLa cell DNA-PK (Fig.  1). As expected, in the absence of the Ku70/80 complex, phosphorylation of the 32-kDa RPA subunit by the DNA-PKcs was much lower than that observed for the holoenzyme (Fig. 1B,  compare lanes b and c). This was likely due to the removal of the majority of the DNA-PKcs from the DNA beads during the washing step. To test whether the RPA kinase assay could discriminate DNA-PK activity from other kinases capable of phosphorylating RPA in cell extracts, we compared the RPA kinase activity in extracts prepared from wild-type and known DNA-PK mutant cell lines. RPA kinase activity was detected in cell extracts derived from the wild-type CHO cell line AA8, but was absent in the CHO K1-derived DNA-PKcs mutant cell line V3 (4) (Fig. 1C, compare lanes b and d). RPA kinase activity was restored to the V3 cell extract by the addition of 50 ng of purified DNA-PK catalytic subunit (Fig. 1C, lane f). Similarly, extracts prepared from the large T antigen immortalized scid mouse cell line SCVA2 (6) also lacked RPA kinase activity in this assay, whereas SCVA2 cells containing human chromosome 8 (SC (8)-10) (29) had strong RPA kinase activity (Fig. 1D,  compare lanes b and d). RPA kinase activity was restored to the SCVA2 cell extracts by addition of purified DNA-PKcs (Fig. 1D, lane f). To determine whether the RPA kinase assay was also sensitive to mutations that disrupt the activity of the Ku complex, we measured RPA phosphorylation using extracts derived from the Ku80 mutant cell line XRS6C and XRS6C cells that were engineered to express the human Ku80 protein (30). We found that RPA kinase activity was absent in the XRS6C cell extracts but was present in extracts derived from the Ku80 expressing XRS6C cells (Fig. 1E, compare lanes b and d).
Since the RPA kinase assay was capable of discriminating between wild-type and known DNA-PK mutant cells, we used it to screen seven DNA double-stranded break repair-deficient rodent cell lines for DNA-PK kinase activity: IRS-20 (23), M10, LX830, SX9, and SX10 (22,23), SL3147, and H5 (24). Of these cells, we found the CHO mutant IRS-20 and the mouse mammary carcinoma cell line SX9 lacked detectable RPA-kinase activity. We chose to first characterize the IRS-20 cell line since it has recently been shown that DNA double-stranded break repair in these cells can be complemented by transfection of human chromosome 8. 2 The human DNA-PKcs gene is located on chromosome 8 (31), suggesting that reduced expression or activity of the DNA-PKcs was responsible for the altered DNA-PK activity in the IRS-20 cell extracts. To determine whether this was true, we compared the DNA-bound RPA kinase activities of extracts prepared from the parental cell line 10B2 and the mutant IRS-20 and IRS-20 cells containing human chromosome 8 (IRS-20(Neo8)). We found that the 32-kDa RPA subunit was phosphorylated in the 10B2 but not the IRS-20 kinase reactions (Fig. 2A, compare lanes b and d). RPA kinase activity was also detected in the IRS-20 (Neo8) kinase reactions (Fig. 2A, lane f) and in reactions performed using IRS-20 cell extracts that had been supplemented with purified DNA-PKcs (Fig. 2A, lane h).
We have shown previously that the CHO cell line V3 displays a severe reduction in the expression of the DNA-PKcs (6). To test whether this was also true for the IRS-20 cells, we measured the levels of both the DNA-PKcs and the Ku70 proteins in the IRS-20 cell extracts by immunoblot analysis using monoclonal antibodies specific for these proteins. Using this assay, we found the amount of DNA-PKcs in the IRS-20 extracts was reduced relative to cell extracts prepared from the parental cell line 10B2 (Fig. 2B, compare lanes a and b). In contrast, we found there was no difference in the abundance of the Ku70 protein when comparing these same cell extracts (Fig. 2B,  compare lanes a and b), which is indicative of the status of both the Ku70 and Ku80 proteins (12,30).
The results of the immunoblot analysis suggested to us that the reduced amount of DNA-PKcs subunit found in the IRS-20 cell extracts could be responsible for the radiosensitive phenotype of these cells. This would be consistent with the defects observed in DNA double-stranded break repair-deficient rodent cells (5,6). However, the level of DNA-PKcs expressed in the IRS-20 cells was much higher than that observed for the CHO V3 cells (Fig. 3A, compare lanes b and d). This indicated to us that in addition to the reduced level of DNA-PKcs expression, the IRS-20 cells might also have defects that affect the assembly of the DNA-PK holoenzyme on DNA or reduce the kinase activity of the enzyme. To test this, we measured the capacity of the DNA-PKcs from the parental 10B2 and the mutant IRS-20 cells to bind to DNA by measuring its retention on linearized DNA covalently linked to agarose beads. Using this assay we found that the amount of DNA-PKcs bound to the DNA-beads was proportional to its abundance in the cell extracts (Fig. 3B, lanes a and b). As a control, we also compared the amount of Ku70 bound to the DNA beads in the same reactions (Fig. 3B, lanes c and d) and found there were no differences between the wild-type 10B2 and IRS-20 mutant cells.
Because the DNA-binding experiments showed the capacity of the IRS-20 DNA-PKcs to bind to the DNA-beads was not significantly different than that of the wild-type enzyme, we tested whether the DNA-PKcs expressed in the IRS-20 cells may have diminished kinase activity relative to the wild-type enzyme. Since the absence of detectable kinase activity in the IRS-20 reactions might reflect limitations in the sensitivity of the RPA phosphorylation assay, we measured the RPA kinase  (XRS6CV, lanes a and b) or transfected with the human Ku80 gene (XRS6C/Ku, lanes c and d). activity in a series of reactions prepared using 10B2 cell extract that was diluted with reaction buffer to give DNA-PK levels equivalent to those found in the IRS-20 reactions. Using this approach, we found that the 10B2 cell extracts contain approximately four times more DNA-PKcs bound to the immobilized DNA than the IRS-20 cell extracts (Fig. 3C). By manipulating the amount of DNA-PKcs in the kinase reactions this way, we found we could detect RPA kinase activity in reactions containing the equivalent of 3 l of 10B2 cell extract (15 g of total protein) (Fig. 3D, lanes a-c). In contrast, we could not detect any RPA kinase activity in the kinase reaction performed in parallel using 50 l of IRS-20 cell extract containing 250 g of total protein (Fig. 3D, lane d). These results indicate that the sensitivity of the RPA kinase assay is sufficient to detect the kinase activity of DNA-PK levels equal to and lower than that found in the IRS-20 extracts and support the idea that the IRS-20 DNA-PKcs has defective kinase activity.
The second cell line that was identified in the initial RPAkinase assay, SX9, was found to have a defect similar to that seen with the IRS-20 cells. RPA kinase activity of cell extracts prepared from the SX9 cells were performed in parallel with extracts prepared from the parental cell line, SR1, and another SR1-derived DNA double-stranded break repair mutant cell line, SX10. Phosphorylation of the 32-kDa RPA subunit was detected in the SR1 and the SX10 cell extracts but was absent in the SX9 cell extract (Fig. 4A). RPA kinase activity of the SX9 extract was not affected by addition of purified human Ku but could be rescued by addition of purified DNA-PKcs (Fig. 4B,  compare lanes b and d). This indicated that the lack of DNA-PK activity in these cells was also due to a defect in the expression or activity of the DNA-PKcs.
To ascertain whether the decrease in DNA-PK activity in the SX9 cell extracts was due to a defect in the expression of the SX9 DNA-PKcs, we probed protein blots of SR1, SX9, and SX10 cell extracts using DNA-PKcs antibodies. The results of this analysis indicated that abundance of the DNA-PKcs in the SX9 cells was reduced relative to both the wild-type SR1 and the SX10 mutant cells (Fig. 4C). However, much like the results obtained with the IRS-20 cells, we could readily detect the DNA-PKcs in the SX-9 cell extracts. Interestingly, the level of DNA-PKcs expressed in the SX-10 cells was slightly reduced relative to the wild-type cells.
To determine if the reduced kinase activity of the SX-9 DNA-PK was due to a defect in the assembly of the enzyme on DNA, we measured the retention of the DNA-PKcs from the SR1, SX-9, and SX-10 cell extracts on DNA-agarose beads. In each case, the level of DNA-PKcs associated with the DNAbeads reflected the relative abundance of the enzyme in whole cell extracts (Fig. 4D). This indicated that the SX9 DNA-PKcs was capable of binding to DNA-agarose beads and suggested that there was no defect in the assembly of the SX9 DNA-PK holoenzyme. Taken together, these results indicate that the SX-9 DNA-PKcs can assemble with Ku to bind to DNA, but has reduced catalytic activity relative to the wild-type enzyme. DISCUSSION The modification of proteins by phosphorylation is a common mechanism for regulating a variety of biochemical activities in eukaryotic cells. Mammalian cell lines displaying defects in the expression of either the catalytic or Ku80 components of DNA-PK have defects in rejoining double-stranded DNA breaks. These results have been interpreted to indicate DNA-PK functions by phosphorylating other proteins involved in these DNA strand rejoining pathways. The data presented in this paper support the hypothesis that the protein kinase activity of DNA-PK is an important component of the doublestranded break rejoining pathway by demonstrating that the DNA repair mutant cell lines IRS-20 and SX9 both express inactive forms of the DNA-PK catalytic subunit. In murine scid and CHO V3 cells, transfer of subgenomic fragments of human chromosome 8 containing the human DNA-PKcs gene rescues the DNA-repair and V(D)J defects and restores expression of a functional DNA-PK holoenzyme (4 -6). Similarly, the radiosensitivity of the IRS-20 cells can also be rescued by human chromosome 8, suggesting that these cells fall into the same genetic complementation group (xrs-7) as the scid and V3 cells (31). Our analysis of the DNA-PK status of the double-stranded break repair mutant SX9 suggests that this cell line is also in the xrs-7 complementation group.
Our data, showing that the DNA-PKcs from both IRS-20 and the SX9 cells can bind to DNA but fails to function as a protein, suggests that the IRS-20 and SX-9 cells harbor mutations that disrupt the kinase activity of the DNA-PKcs. This could be mediated via changes in critical residues in the kinase active site or by mutations that disrupt the global structure of the enzyme. Such mutations might affect both the activity of the enzyme as well as its intracellular stability. Interestingly, the levels of DNA-PKcs found in both the IRS-20 and SX9 cells is reduced relative to their parental, wild-type cell lines. This also occurs in both the murine scid and CHO V3 cells where the expression of the DNA-PKcs is severely repressed relative to wild-type levels and there is no apparent DNA-PK activity (4 -6). Recently, it was shown that scid mice contain a mutation in the C terminus of the DNA-PKcs gene corresponding to amino acid position 4045 that results in the introduction of a stop codon (32). This mutation may disrupt the structure of the phosphatidylinositol-3 kinase domain of DNA-PKcs, thereby inhibiting the kinase activity of the enzyme. Furthermore, it is suggested this mutation may also account for the decreased expression of DNA-PKcs observed in scid cells by reducing the stability of the gene product.
Despite the detailed information on how DNA-PK functions in vitro, it is not clear how this enzyme participates in DNA double-stranded break rejoining pathways. Subsequent to the generation of a double-stranded DNA break due to exposure of a cell to a strand-breaking agent or during the course of V(D)J recombination, DNA-PK would likely bind to the ends of the broken DNA, becoming colocalized with components of the strand-rejoining machinery. The phosphorylation of some of these colocalized repair factors might serve to activate some other biochemical activity such as DNA-unwinding or ligation. However, it is also possible that in addition to serving as an activator of repair, DNA-PK could function to stimulate the dissociation of other proteins bound nearby the DNA-ends that prevent accessibility of the repair machinery.
In both of these models, the kinase activity of DNA-PK is a key component of the strand-rejoining reaction. A central question in the study of the DNA-repair defects in cells mutated in the Ku or DNA-PKcs genes is how the inactivation of the DNA-PK complex alters the repair process. In both the Ku and DNA-PKcs mutant cells, the overall level of double-stranded DNA break repair is nearly 50% that of the wild-type cells (33)(34)(35). This may indicate that DNA-PK-independent repair pathways are capable of rejoining some populations of the double-stranded DNA breaks and that DNA-PK is only required for the repair of the remaining subset of doublestranded breaks (36). This subset of damaged DNA may require the activity of DNA-PK for strand break rejoining because of the nature of the damaged DNA, the chromosomal location, or chromatin context of the damaged sequence.